arrow_schema/
schema.rs

1// Licensed to the Apache Software Foundation (ASF) under one
2// or more contributor license agreements.  See the NOTICE file
3// distributed with this work for additional information
4// regarding copyright ownership.  The ASF licenses this file
5// to you under the Apache License, Version 2.0 (the
6// "License"); you may not use this file except in compliance
7// with the License.  You may obtain a copy of the License at
8//
9//   http://www.apache.org/licenses/LICENSE-2.0
10//
11// Unless required by applicable law or agreed to in writing,
12// software distributed under the License is distributed on an
13// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
14// KIND, either express or implied.  See the License for the
15// specific language governing permissions and limitations
16// under the License.
17
18use std::collections::HashMap;
19use std::fmt;
20use std::hash::Hash;
21use std::sync::Arc;
22
23use crate::error::ArrowError;
24use crate::field::Field;
25use crate::{DataType, FieldRef, Fields};
26
27/// A builder to facilitate building a [`Schema`] from iteratively from [`FieldRef`]
28#[derive(Debug, Default)]
29pub struct SchemaBuilder {
30    fields: Vec<FieldRef>,
31    metadata: HashMap<String, String>,
32}
33
34impl SchemaBuilder {
35    /// Creates a new empty [`SchemaBuilder`]
36    pub fn new() -> Self {
37        Self::default()
38    }
39
40    /// Creates a new empty [`SchemaBuilder`] with space for `capacity` fields
41    pub fn with_capacity(capacity: usize) -> Self {
42        Self {
43            fields: Vec::with_capacity(capacity),
44            metadata: Default::default(),
45        }
46    }
47
48    /// Appends a [`FieldRef`] to this [`SchemaBuilder`] without checking for collision
49    pub fn push(&mut self, field: impl Into<FieldRef>) {
50        self.fields.push(field.into())
51    }
52
53    /// Removes and returns the [`FieldRef`] as index `idx`
54    ///
55    /// # Panics
56    ///
57    /// Panics if index out of bounds
58    pub fn remove(&mut self, idx: usize) -> FieldRef {
59        self.fields.remove(idx)
60    }
61
62    /// Returns an immutable reference to the [`FieldRef`] at index `idx`
63    ///
64    /// # Panics
65    ///
66    /// Panics if index out of bounds
67    pub fn field(&mut self, idx: usize) -> &FieldRef {
68        &mut self.fields[idx]
69    }
70
71    /// Returns a mutable reference to the [`FieldRef`] at index `idx`
72    ///
73    /// # Panics
74    ///
75    /// Panics if index out of bounds
76    pub fn field_mut(&mut self, idx: usize) -> &mut FieldRef {
77        &mut self.fields[idx]
78    }
79
80    /// Returns an immutable reference to the Map of custom metadata key-value pairs.
81    pub fn metadata(&mut self) -> &HashMap<String, String> {
82        &self.metadata
83    }
84
85    /// Returns a mutable reference to the Map of custom metadata key-value pairs.
86    pub fn metadata_mut(&mut self) -> &mut HashMap<String, String> {
87        &mut self.metadata
88    }
89
90    /// Reverse the fileds
91    pub fn reverse(&mut self) {
92        self.fields.reverse();
93    }
94
95    /// Appends a [`FieldRef`] to this [`SchemaBuilder`] checking for collision
96    ///
97    /// If an existing field exists with the same name, calls [`Field::try_merge`]
98    pub fn try_merge(&mut self, field: &FieldRef) -> Result<(), ArrowError> {
99        // This could potentially be sped up with a HashMap or similar
100        let existing = self.fields.iter_mut().find(|f| f.name() == field.name());
101        match existing {
102            Some(e) if Arc::ptr_eq(e, field) => {} // Nothing to do
103            Some(e) => match Arc::get_mut(e) {
104                Some(e) => e.try_merge(field.as_ref())?,
105                None => {
106                    let mut t = e.as_ref().clone();
107                    t.try_merge(field)?;
108                    *e = Arc::new(t)
109                }
110            },
111            None => self.fields.push(field.clone()),
112        }
113        Ok(())
114    }
115
116    /// Consume this [`SchemaBuilder`] yielding the final [`Schema`]
117    pub fn finish(self) -> Schema {
118        Schema {
119            fields: self.fields.into(),
120            metadata: self.metadata,
121        }
122    }
123}
124
125impl From<&Fields> for SchemaBuilder {
126    fn from(value: &Fields) -> Self {
127        Self {
128            fields: value.to_vec(),
129            metadata: Default::default(),
130        }
131    }
132}
133
134impl From<Fields> for SchemaBuilder {
135    fn from(value: Fields) -> Self {
136        Self {
137            fields: value.to_vec(),
138            metadata: Default::default(),
139        }
140    }
141}
142
143impl From<&Schema> for SchemaBuilder {
144    fn from(value: &Schema) -> Self {
145        Self::from(value.clone())
146    }
147}
148
149impl From<Schema> for SchemaBuilder {
150    fn from(value: Schema) -> Self {
151        Self {
152            fields: value.fields.to_vec(),
153            metadata: value.metadata,
154        }
155    }
156}
157
158impl Extend<FieldRef> for SchemaBuilder {
159    fn extend<T: IntoIterator<Item = FieldRef>>(&mut self, iter: T) {
160        let iter = iter.into_iter();
161        self.fields.reserve(iter.size_hint().0);
162        for f in iter {
163            self.push(f)
164        }
165    }
166}
167
168impl Extend<Field> for SchemaBuilder {
169    fn extend<T: IntoIterator<Item = Field>>(&mut self, iter: T) {
170        let iter = iter.into_iter();
171        self.fields.reserve(iter.size_hint().0);
172        for f in iter {
173            self.push(f)
174        }
175    }
176}
177
178/// A reference-counted reference to a [`Schema`].
179pub type SchemaRef = Arc<Schema>;
180
181/// Describes the meta-data of an ordered sequence of relative types.
182///
183/// Note that this information is only part of the meta-data and not part of the physical
184/// memory layout.
185#[derive(Debug, Clone, PartialEq, Eq)]
186#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
187pub struct Schema {
188    /// A sequence of fields that describe the schema.
189    pub fields: Fields,
190    /// A map of key-value pairs containing additional metadata.
191    pub metadata: HashMap<String, String>,
192}
193
194impl Schema {
195    /// Creates an empty `Schema`
196    pub fn empty() -> Self {
197        Self {
198            fields: Default::default(),
199            metadata: HashMap::new(),
200        }
201    }
202
203    /// Creates a new [`Schema`] from a sequence of [`Field`] values.
204    ///
205    /// # Example
206    ///
207    /// ```
208    /// # use arrow_schema::*;
209    /// let field_a = Field::new("a", DataType::Int64, false);
210    /// let field_b = Field::new("b", DataType::Boolean, false);
211    ///
212    /// let schema = Schema::new(vec![field_a, field_b]);
213    /// ```
214    pub fn new(fields: impl Into<Fields>) -> Self {
215        Self::new_with_metadata(fields, HashMap::new())
216    }
217
218    /// Creates a new [`Schema`] from a sequence of [`Field`] values
219    /// and adds additional metadata in form of key value pairs.
220    ///
221    /// # Example
222    ///
223    /// ```
224    /// # use arrow_schema::*;
225    /// # use std::collections::HashMap;
226    ///
227    /// let field_a = Field::new("a", DataType::Int64, false);
228    /// let field_b = Field::new("b", DataType::Boolean, false);
229    ///
230    /// let mut metadata: HashMap<String, String> = HashMap::new();
231    /// metadata.insert("row_count".to_string(), "100".to_string());
232    ///
233    /// let schema = Schema::new_with_metadata(vec![field_a, field_b], metadata);
234    /// ```
235    #[inline]
236    pub fn new_with_metadata(fields: impl Into<Fields>, metadata: HashMap<String, String>) -> Self {
237        Self {
238            fields: fields.into(),
239            metadata,
240        }
241    }
242
243    /// Sets the metadata of this `Schema` to be `metadata` and returns self
244    pub fn with_metadata(mut self, metadata: HashMap<String, String>) -> Self {
245        self.metadata = metadata;
246        self
247    }
248
249    /// Returns a new schema with only the specified columns in the new schema
250    /// This carries metadata from the parent schema over as well
251    pub fn project(&self, indices: &[usize]) -> Result<Schema, ArrowError> {
252        let new_fields = indices
253            .iter()
254            .map(|i| {
255                self.fields.get(*i).cloned().ok_or_else(|| {
256                    ArrowError::SchemaError(format!(
257                        "project index {} out of bounds, max field {}",
258                        i,
259                        self.fields().len()
260                    ))
261                })
262            })
263            .collect::<Result<Vec<_>, _>>()?;
264        Ok(Self::new_with_metadata(new_fields, self.metadata.clone()))
265    }
266
267    /// Merge schema into self if it is compatible. Struct fields will be merged recursively.
268    ///
269    /// Example:
270    ///
271    /// ```
272    /// # use arrow_schema::*;
273    ///
274    /// let merged = Schema::try_merge(vec![
275    ///     Schema::new(vec![
276    ///         Field::new("c1", DataType::Int64, false),
277    ///         Field::new("c2", DataType::Utf8, false),
278    ///     ]),
279    ///     Schema::new(vec![
280    ///         Field::new("c1", DataType::Int64, true),
281    ///         Field::new("c2", DataType::Utf8, false),
282    ///         Field::new("c3", DataType::Utf8, false),
283    ///     ]),
284    /// ]).unwrap();
285    ///
286    /// assert_eq!(
287    ///     merged,
288    ///     Schema::new(vec![
289    ///         Field::new("c1", DataType::Int64, true),
290    ///         Field::new("c2", DataType::Utf8, false),
291    ///         Field::new("c3", DataType::Utf8, false),
292    ///     ]),
293    /// );
294    /// ```
295    pub fn try_merge(schemas: impl IntoIterator<Item = Self>) -> Result<Self, ArrowError> {
296        let mut out_meta = HashMap::new();
297        let mut out_fields = SchemaBuilder::new();
298        for schema in schemas {
299            let Schema { metadata, fields } = schema;
300
301            // merge metadata
302            for (key, value) in metadata.into_iter() {
303                if let Some(old_val) = out_meta.get(&key) {
304                    if old_val != &value {
305                        return Err(ArrowError::SchemaError(format!(
306                            "Fail to merge schema due to conflicting metadata. \
307                                         Key '{key}' has different values '{old_val}' and '{value}'"
308                        )));
309                    }
310                }
311                out_meta.insert(key, value);
312            }
313
314            // merge fields
315            fields.iter().try_for_each(|x| out_fields.try_merge(x))?
316        }
317
318        Ok(out_fields.finish().with_metadata(out_meta))
319    }
320
321    /// Returns an immutable reference of the vector of `Field` instances.
322    #[inline]
323    pub const fn fields(&self) -> &Fields {
324        &self.fields
325    }
326
327    /// Returns a vector with references to all fields (including nested fields)
328    ///
329    /// # Example
330    ///
331    /// ```
332    /// use std::sync::Arc;
333    /// use arrow_schema::{DataType, Field, Fields, Schema};
334    ///
335    /// let f1 = Arc::new(Field::new("a", DataType::Boolean, false));
336    ///
337    /// let f2_inner = Arc::new(Field::new("b_inner", DataType::Int8, false));
338    /// let f2 = Arc::new(Field::new("b", DataType::List(f2_inner.clone()), false));
339    ///
340    /// let f3_inner1 = Arc::new(Field::new("c_inner1", DataType::Int8, false));
341    /// let f3_inner2 = Arc::new(Field::new("c_inner2", DataType::Int8, false));
342    /// let f3 = Arc::new(Field::new(
343    ///     "c",
344    ///     DataType::Struct(vec![f3_inner1.clone(), f3_inner2.clone()].into()),
345    ///     false
346    /// ));
347    ///
348    /// let mut schema = Schema::new(vec![
349    ///   f1.clone(), f2.clone(), f3.clone()
350    /// ]);
351    /// assert_eq!(
352    ///     schema.flattened_fields(),
353    ///     vec![
354    ///         f1.as_ref(),
355    ///         f2.as_ref(),
356    ///         f2_inner.as_ref(),
357    ///         f3.as_ref(),
358    ///         f3_inner1.as_ref(),
359    ///         f3_inner2.as_ref()
360    ///    ]
361    /// );
362    /// ```
363    #[inline]
364    pub fn flattened_fields(&self) -> Vec<&Field> {
365        self.fields.iter().flat_map(|f| f.fields()).collect()
366    }
367
368    /// Returns an immutable reference of a specific [`Field`] instance selected using an
369    /// offset within the internal `fields` vector.
370    ///
371    /// # Panics
372    ///
373    /// Panics if index out of bounds
374    pub fn field(&self, i: usize) -> &Field {
375        &self.fields[i]
376    }
377
378    /// Returns an immutable reference of a specific [`Field`] instance selected by name.
379    pub fn field_with_name(&self, name: &str) -> Result<&Field, ArrowError> {
380        Ok(&self.fields[self.index_of(name)?])
381    }
382
383    /// Returns a vector of immutable references to all [`Field`] instances selected by
384    /// the dictionary ID they use.
385    #[deprecated(
386        since = "54.0.0",
387        note = "The ability to preserve dictionary IDs will be removed. With it, all functions related to it."
388    )]
389    pub fn fields_with_dict_id(&self, dict_id: i64) -> Vec<&Field> {
390        #[allow(deprecated)]
391        self.fields
392            .iter()
393            .flat_map(|f| f.fields_with_dict_id(dict_id))
394            .collect()
395    }
396
397    /// Find the index of the column with the given name.
398    pub fn index_of(&self, name: &str) -> Result<usize, ArrowError> {
399        let (idx, _) = self.fields().find(name).ok_or_else(|| {
400            let valid_fields: Vec<_> = self.fields.iter().map(|f| f.name()).collect();
401            ArrowError::SchemaError(format!(
402                "Unable to get field named \"{name}\". Valid fields: {valid_fields:?}"
403            ))
404        })?;
405        Ok(idx)
406    }
407
408    /// Returns an immutable reference to the Map of custom metadata key-value pairs.
409    #[inline]
410    pub const fn metadata(&self) -> &HashMap<String, String> {
411        &self.metadata
412    }
413
414    /// Normalize a [`Schema`] into a flat table.
415    ///
416    /// Nested [`Field`]s will generate names separated by `separator`, up to a depth of `max_level`
417    /// (unlimited if `None`).
418    ///
419    /// e.g. given a [`Schema`]:
420    ///
421    /// ```text
422    ///     "foo": StructArray<"bar": Utf8>
423    /// ```
424    ///
425    /// A separator of `"."` would generate a batch with the schema:
426    ///
427    /// ```text
428    ///     "foo.bar": Utf8
429    /// ```
430    ///
431    /// Note that giving a depth of `Some(0)` to `max_level` is the same as passing in `None`;
432    /// it will be treated as unlimited.
433    ///
434    /// # Example
435    ///
436    /// ```
437    /// # use std::sync::Arc;
438    /// # use arrow_schema::{DataType, Field, Fields, Schema};
439    /// let schema = Schema::new(vec![
440    ///     Field::new(
441    ///         "a",
442    ///         DataType::Struct(Fields::from(vec![
443    ///             Arc::new(Field::new("animals", DataType::Utf8, true)),
444    ///             Arc::new(Field::new("n_legs", DataType::Int64, true)),
445    ///         ])),
446    ///         false,
447    ///     ),
448    /// ])
449    /// .normalize(".", None)
450    /// .expect("valid normalization");
451    /// let expected = Schema::new(vec![
452    ///     Field::new("a.animals", DataType::Utf8, true),
453    ///     Field::new("a.n_legs", DataType::Int64, true),
454    /// ]);
455    /// assert_eq!(schema, expected);
456    /// ```
457    pub fn normalize(&self, separator: &str, max_level: Option<usize>) -> Result<Self, ArrowError> {
458        let max_level = match max_level.unwrap_or(usize::MAX) {
459            0 => usize::MAX,
460            val => val,
461        };
462        let mut stack: Vec<(usize, Vec<&str>, &FieldRef)> = self
463            .fields()
464            .iter()
465            .rev()
466            .map(|f| {
467                let name_vec: Vec<&str> = vec![f.name()];
468                (0, name_vec, f)
469            })
470            .collect();
471        let mut fields: Vec<FieldRef> = Vec::new();
472
473        while let Some((depth, name, field_ref)) = stack.pop() {
474            match field_ref.data_type() {
475                DataType::Struct(ff) if depth < max_level => {
476                    // Need to zip these in reverse to maintain original order
477                    for fff in ff.into_iter().rev() {
478                        let mut name = name.clone();
479                        name.push(separator);
480                        name.push(fff.name());
481                        stack.push((depth + 1, name, fff))
482                    }
483                }
484                _ => {
485                    let updated_field = Field::new(
486                        name.concat(),
487                        field_ref.data_type().clone(),
488                        field_ref.is_nullable(),
489                    );
490                    fields.push(Arc::new(updated_field));
491                }
492            }
493        }
494        Ok(Schema::new(fields))
495    }
496
497    /// Look up a column by name and return a immutable reference to the column along with
498    /// its index.
499    pub fn column_with_name(&self, name: &str) -> Option<(usize, &Field)> {
500        let (idx, field) = self.fields.find(name)?;
501        Some((idx, field.as_ref()))
502    }
503
504    /// Check to see if `self` is a superset of `other` schema.
505    ///
506    /// In particular returns true if `self.metadata` is a superset of `other.metadata`
507    /// and [`Fields::contains`] for `self.fields` and `other.fields`
508    ///
509    /// In other words, any record that conforms to `other` should also conform to `self`.
510    pub fn contains(&self, other: &Schema) -> bool {
511        // make sure self.metadata is a superset of other.metadata
512        self.fields.contains(&other.fields)
513            && other
514                .metadata
515                .iter()
516                .all(|(k, v1)| self.metadata.get(k).map(|v2| v1 == v2).unwrap_or_default())
517    }
518}
519
520impl fmt::Display for Schema {
521    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
522        f.write_str(
523            &self
524                .fields
525                .iter()
526                .map(|c| c.to_string())
527                .collect::<Vec<String>>()
528                .join(", "),
529        )
530    }
531}
532
533// need to implement `Hash` manually because `HashMap` implement Eq but no `Hash`
534#[allow(clippy::derived_hash_with_manual_eq)]
535impl Hash for Schema {
536    fn hash<H: std::hash::Hasher>(&self, state: &mut H) {
537        self.fields.hash(state);
538
539        // ensure deterministic key order
540        let mut keys: Vec<&String> = self.metadata.keys().collect();
541        keys.sort();
542        for k in keys {
543            k.hash(state);
544            self.metadata.get(k).expect("key valid").hash(state);
545        }
546    }
547}
548
549#[cfg(test)]
550mod tests {
551    use crate::datatype::DataType;
552    use crate::{TimeUnit, UnionMode};
553
554    use super::*;
555
556    #[test]
557    #[cfg(feature = "serde")]
558    fn test_ser_de_metadata() {
559        // ser/de with empty metadata
560        let schema = Schema::new(vec![
561            Field::new("name", DataType::Utf8, false),
562            Field::new("address", DataType::Utf8, false),
563            Field::new("priority", DataType::UInt8, false),
564        ]);
565
566        let json = serde_json::to_string(&schema).unwrap();
567        let de_schema = serde_json::from_str(&json).unwrap();
568
569        assert_eq!(schema, de_schema);
570
571        // ser/de with non-empty metadata
572        let schema =
573            schema.with_metadata([("key".to_owned(), "val".to_owned())].into_iter().collect());
574        let json = serde_json::to_string(&schema).unwrap();
575        let de_schema = serde_json::from_str(&json).unwrap();
576
577        assert_eq!(schema, de_schema);
578    }
579
580    #[test]
581    fn test_projection() {
582        let mut metadata = HashMap::new();
583        metadata.insert("meta".to_string(), "data".to_string());
584
585        let schema = Schema::new(vec![
586            Field::new("name", DataType::Utf8, false),
587            Field::new("address", DataType::Utf8, false),
588            Field::new("priority", DataType::UInt8, false),
589        ])
590        .with_metadata(metadata);
591
592        let projected: Schema = schema.project(&[0, 2]).unwrap();
593
594        assert_eq!(projected.fields().len(), 2);
595        assert_eq!(projected.fields()[0].name(), "name");
596        assert_eq!(projected.fields()[1].name(), "priority");
597        assert_eq!(projected.metadata.get("meta").unwrap(), "data")
598    }
599
600    #[test]
601    fn test_oob_projection() {
602        let mut metadata = HashMap::new();
603        metadata.insert("meta".to_string(), "data".to_string());
604
605        let schema = Schema::new(vec![
606            Field::new("name", DataType::Utf8, false),
607            Field::new("address", DataType::Utf8, false),
608            Field::new("priority", DataType::UInt8, false),
609        ])
610        .with_metadata(metadata);
611
612        let projected = schema.project(&[0, 3]);
613
614        assert!(projected.is_err());
615        if let Err(e) = projected {
616            assert_eq!(
617                e.to_string(),
618                "Schema error: project index 3 out of bounds, max field 3".to_string()
619            )
620        }
621    }
622
623    #[test]
624    fn test_schema_contains() {
625        let mut metadata1 = HashMap::new();
626        metadata1.insert("meta".to_string(), "data".to_string());
627
628        let schema1 = Schema::new(vec![
629            Field::new("name", DataType::Utf8, false),
630            Field::new("address", DataType::Utf8, false),
631            Field::new("priority", DataType::UInt8, false),
632        ])
633        .with_metadata(metadata1.clone());
634
635        let mut metadata2 = HashMap::new();
636        metadata2.insert("meta".to_string(), "data".to_string());
637        metadata2.insert("meta2".to_string(), "data".to_string());
638        let schema2 = Schema::new(vec![
639            Field::new("name", DataType::Utf8, false),
640            Field::new("address", DataType::Utf8, false),
641            Field::new("priority", DataType::UInt8, false),
642        ])
643        .with_metadata(metadata2);
644
645        // reflexivity
646        assert!(schema1.contains(&schema1));
647        assert!(schema2.contains(&schema2));
648
649        assert!(!schema1.contains(&schema2));
650        assert!(schema2.contains(&schema1));
651    }
652
653    #[test]
654    fn schema_equality() {
655        let schema1 = Schema::new(vec![
656            Field::new("c1", DataType::Utf8, false),
657            Field::new("c2", DataType::Float64, true),
658            Field::new("c3", DataType::LargeBinary, true),
659        ]);
660        let schema2 = Schema::new(vec![
661            Field::new("c1", DataType::Utf8, false),
662            Field::new("c2", DataType::Float64, true),
663            Field::new("c3", DataType::LargeBinary, true),
664        ]);
665
666        assert_eq!(schema1, schema2);
667
668        let schema3 = Schema::new(vec![
669            Field::new("c1", DataType::Utf8, false),
670            Field::new("c2", DataType::Float32, true),
671        ]);
672        let schema4 = Schema::new(vec![
673            Field::new("C1", DataType::Utf8, false),
674            Field::new("C2", DataType::Float64, true),
675        ]);
676
677        assert_ne!(schema1, schema3);
678        assert_ne!(schema1, schema4);
679        assert_ne!(schema2, schema3);
680        assert_ne!(schema2, schema4);
681        assert_ne!(schema3, schema4);
682
683        let f = Field::new("c1", DataType::Utf8, false).with_metadata(
684            [("foo".to_string(), "bar".to_string())]
685                .iter()
686                .cloned()
687                .collect(),
688        );
689        let schema5 = Schema::new(vec![
690            f,
691            Field::new("c2", DataType::Float64, true),
692            Field::new("c3", DataType::LargeBinary, true),
693        ]);
694        assert_ne!(schema1, schema5);
695    }
696
697    #[test]
698    fn create_schema_string() {
699        let schema = person_schema();
700        assert_eq!(
701            schema.to_string(),
702            "Field { \"first_name\": Utf8, metadata: {\"k\": \"v\"} }, \
703             Field { \"last_name\": Utf8 }, \
704             Field { \"address\": Struct(\"street\": Utf8, \"zip\": UInt16) }, \
705             Field { \"interests\": nullable Dictionary(Int32, Utf8), dict_id: 123, dict_is_ordered }"
706        )
707    }
708
709    #[test]
710    fn schema_field_accessors() {
711        let schema = person_schema();
712
713        // test schema accessors
714        assert_eq!(schema.fields().len(), 4);
715
716        // test field accessors
717        let first_name = &schema.fields()[0];
718        assert_eq!(first_name.name(), "first_name");
719        assert_eq!(first_name.data_type(), &DataType::Utf8);
720        assert!(!first_name.is_nullable());
721        #[allow(deprecated)]
722        let dict_id = first_name.dict_id();
723        assert_eq!(dict_id, None);
724        assert_eq!(first_name.dict_is_ordered(), None);
725
726        let metadata = first_name.metadata();
727        assert!(!metadata.is_empty());
728        let md = &metadata;
729        assert_eq!(md.len(), 1);
730        let key = md.get("k");
731        assert!(key.is_some());
732        assert_eq!(key.unwrap(), "v");
733
734        let interests = &schema.fields()[3];
735        assert_eq!(interests.name(), "interests");
736        assert_eq!(
737            interests.data_type(),
738            &DataType::Dictionary(Box::new(DataType::Int32), Box::new(DataType::Utf8))
739        );
740        #[allow(deprecated)]
741        let dict_id = interests.dict_id();
742        assert_eq!(dict_id, Some(123));
743        assert_eq!(interests.dict_is_ordered(), Some(true));
744    }
745
746    #[test]
747    #[should_panic(
748        expected = "Unable to get field named \\\"nickname\\\". Valid fields: [\\\"first_name\\\", \\\"last_name\\\", \\\"address\\\", \\\"interests\\\"]"
749    )]
750    fn schema_index_of() {
751        let schema = person_schema();
752        assert_eq!(schema.index_of("first_name").unwrap(), 0);
753        assert_eq!(schema.index_of("last_name").unwrap(), 1);
754        schema.index_of("nickname").unwrap();
755    }
756
757    #[test]
758    fn normalize_simple() {
759        let schema = Schema::new(vec![
760            Field::new(
761                "a",
762                DataType::Struct(Fields::from(vec![
763                    Arc::new(Field::new("animals", DataType::Utf8, true)),
764                    Arc::new(Field::new("n_legs", DataType::Int64, true)),
765                    Arc::new(Field::new("year", DataType::Int64, true)),
766                ])),
767                false,
768            ),
769            Field::new("month", DataType::Int64, true),
770        ])
771        .normalize(".", Some(0))
772        .expect("valid normalization");
773
774        let expected = Schema::new(vec![
775            Field::new("a.animals", DataType::Utf8, true),
776            Field::new("a.n_legs", DataType::Int64, true),
777            Field::new("a.year", DataType::Int64, true),
778            Field::new("month", DataType::Int64, true),
779        ]);
780
781        assert_eq!(schema, expected);
782
783        // Check that 0, None have the same result
784        let schema = Schema::new(vec![
785            Field::new(
786                "a",
787                DataType::Struct(Fields::from(vec![
788                    Arc::new(Field::new("animals", DataType::Utf8, true)),
789                    Arc::new(Field::new("n_legs", DataType::Int64, true)),
790                    Arc::new(Field::new("year", DataType::Int64, true)),
791                ])),
792                false,
793            ),
794            Field::new("month", DataType::Int64, true),
795        ])
796        .normalize(".", None)
797        .expect("valid normalization");
798
799        assert_eq!(schema, expected);
800    }
801
802    #[test]
803    fn normalize_nested() {
804        let a = Arc::new(Field::new("a", DataType::Utf8, true));
805        let b = Arc::new(Field::new("b", DataType::Int64, false));
806        let c = Arc::new(Field::new("c", DataType::Int64, true));
807
808        let d = Arc::new(Field::new("d", DataType::Utf8, true));
809        let e = Arc::new(Field::new("e", DataType::Int64, false));
810        let f = Arc::new(Field::new("f", DataType::Int64, true));
811
812        let one = Arc::new(Field::new(
813            "1",
814            DataType::Struct(Fields::from(vec![a.clone(), b.clone(), c.clone()])),
815            false,
816        ));
817        let two = Arc::new(Field::new(
818            "2",
819            DataType::Struct(Fields::from(vec![d.clone(), e.clone(), f.clone()])),
820            true,
821        ));
822
823        let exclamation = Arc::new(Field::new(
824            "!",
825            DataType::Struct(Fields::from(vec![one, two])),
826            false,
827        ));
828
829        let normalize_all = Schema::new(vec![exclamation.clone()])
830            .normalize(".", Some(0))
831            .expect("valid normalization");
832
833        let expected = Schema::new(vec![
834            Field::new("!.1.a", DataType::Utf8, true),
835            Field::new("!.1.b", DataType::Int64, false),
836            Field::new("!.1.c", DataType::Int64, true),
837            Field::new("!.2.d", DataType::Utf8, true),
838            Field::new("!.2.e", DataType::Int64, false),
839            Field::new("!.2.f", DataType::Int64, true),
840        ]);
841
842        assert_eq!(normalize_all, expected);
843
844        let normalize_depth_one = Schema::new(vec![exclamation])
845            .normalize(".", Some(1))
846            .expect("valid normalization");
847
848        let expected = Schema::new(vec![
849            Field::new("!.1", DataType::Struct(Fields::from(vec![a, b, c])), false),
850            Field::new("!.2", DataType::Struct(Fields::from(vec![d, e, f])), true),
851        ]);
852
853        assert_eq!(normalize_depth_one, expected);
854    }
855
856    #[test]
857    fn normalize_list() {
858        // Only the Struct type field should be unwrapped
859        let a = Arc::new(Field::new("a", DataType::Utf8, true));
860        let b = Arc::new(Field::new("b", DataType::Int64, false));
861        let c = Arc::new(Field::new("c", DataType::Int64, true));
862        let d = Arc::new(Field::new("d", DataType::Utf8, true));
863        let e = Arc::new(Field::new("e", DataType::Int64, false));
864        let f = Arc::new(Field::new("f", DataType::Int64, true));
865
866        let one = Arc::new(Field::new(
867            "1",
868            DataType::Struct(Fields::from(vec![a.clone(), b.clone(), c.clone()])),
869            true,
870        ));
871
872        let two = Arc::new(Field::new(
873            "2",
874            DataType::List(Arc::new(Field::new_list_field(
875                DataType::Struct(Fields::from(vec![d.clone(), e.clone(), f.clone()])),
876                true,
877            ))),
878            false,
879        ));
880
881        let exclamation = Arc::new(Field::new(
882            "!",
883            DataType::Struct(Fields::from(vec![one.clone(), two.clone()])),
884            false,
885        ));
886
887        let normalize_all = Schema::new(vec![exclamation.clone()])
888            .normalize(".", None)
889            .expect("valid normalization");
890
891        // List shouldn't be affected
892        let expected = Schema::new(vec![
893            Field::new("!.1.a", DataType::Utf8, true),
894            Field::new("!.1.b", DataType::Int64, false),
895            Field::new("!.1.c", DataType::Int64, true),
896            Field::new(
897                "!.2",
898                DataType::List(Arc::new(Field::new_list_field(
899                    DataType::Struct(Fields::from(vec![d.clone(), e.clone(), f.clone()])),
900                    true,
901                ))),
902                false,
903            ),
904        ]);
905
906        assert_eq!(normalize_all, expected);
907        assert_eq!(normalize_all.fields().len(), 4);
908
909        // FixedSizeList
910        let two = Arc::new(Field::new(
911            "2",
912            DataType::FixedSizeList(
913                Arc::new(Field::new_fixed_size_list(
914                    "3",
915                    Arc::new(Field::new_list_field(
916                        DataType::Struct(Fields::from(vec![d.clone(), e.clone(), f.clone()])),
917                        true,
918                    )),
919                    1,
920                    true,
921                )),
922                1,
923            ),
924            false,
925        ));
926
927        let exclamation = Arc::new(Field::new(
928            "!",
929            DataType::Struct(Fields::from(vec![one.clone(), two])),
930            false,
931        ));
932
933        let normalize_all = Schema::new(vec![exclamation.clone()])
934            .normalize(".", None)
935            .expect("valid normalization");
936
937        // FixedSizeList shouldn't be affected
938        let expected = Schema::new(vec![
939            Field::new("!.1.a", DataType::Utf8, true),
940            Field::new("!.1.b", DataType::Int64, false),
941            Field::new("!.1.c", DataType::Int64, true),
942            Field::new(
943                "!.2",
944                DataType::FixedSizeList(
945                    Arc::new(Field::new_fixed_size_list(
946                        "3",
947                        Arc::new(Field::new_list_field(
948                            DataType::Struct(Fields::from(vec![d.clone(), e.clone(), f.clone()])),
949                            true,
950                        )),
951                        1,
952                        true,
953                    )),
954                    1,
955                ),
956                false,
957            ),
958        ]);
959
960        assert_eq!(normalize_all, expected);
961        assert_eq!(normalize_all.fields().len(), 4);
962
963        // LargeList
964        let two = Arc::new(Field::new(
965            "2",
966            DataType::FixedSizeList(
967                Arc::new(Field::new_large_list(
968                    "3",
969                    Arc::new(Field::new_list_field(
970                        DataType::Struct(Fields::from(vec![d.clone(), e.clone(), f.clone()])),
971                        true,
972                    )),
973                    true,
974                )),
975                1,
976            ),
977            false,
978        ));
979
980        let exclamation = Arc::new(Field::new(
981            "!",
982            DataType::Struct(Fields::from(vec![one.clone(), two])),
983            false,
984        ));
985
986        let normalize_all = Schema::new(vec![exclamation.clone()])
987            .normalize(".", None)
988            .expect("valid normalization");
989
990        // LargeList shouldn't be affected
991        let expected = Schema::new(vec![
992            Field::new("!.1.a", DataType::Utf8, true),
993            Field::new("!.1.b", DataType::Int64, false),
994            Field::new("!.1.c", DataType::Int64, true),
995            Field::new(
996                "!.2",
997                DataType::FixedSizeList(
998                    Arc::new(Field::new_large_list(
999                        "3",
1000                        Arc::new(Field::new_list_field(
1001                            DataType::Struct(Fields::from(vec![d.clone(), e.clone(), f.clone()])),
1002                            true,
1003                        )),
1004                        true,
1005                    )),
1006                    1,
1007                ),
1008                false,
1009            ),
1010        ]);
1011
1012        assert_eq!(normalize_all, expected);
1013        assert_eq!(normalize_all.fields().len(), 4);
1014    }
1015
1016    #[test]
1017    fn normalize_deep_nested() {
1018        // No unwrapping expected
1019        let a = Arc::new(Field::new("a", DataType::Utf8, true));
1020        let b = Arc::new(Field::new("b", DataType::Int64, false));
1021        let c = Arc::new(Field::new("c", DataType::Int64, true));
1022        let d = Arc::new(Field::new("d", DataType::Utf8, true));
1023        let e = Arc::new(Field::new("e", DataType::Int64, false));
1024        let f = Arc::new(Field::new("f", DataType::Int64, true));
1025
1026        let one = Arc::new(Field::new(
1027            "1",
1028            DataType::Struct(Fields::from(vec![a.clone(), b.clone(), c.clone()])),
1029            true,
1030        ));
1031
1032        let two = Arc::new(Field::new(
1033            "2",
1034            DataType::List(Arc::new(Field::new_list_field(
1035                DataType::Struct(Fields::from(vec![d.clone(), e.clone(), f.clone()])),
1036                true,
1037            ))),
1038            false,
1039        ));
1040
1041        let l10 = Arc::new(Field::new(
1042            "l10",
1043            DataType::List(Arc::new(Field::new_list_field(
1044                DataType::Struct(Fields::from(vec![one, two])),
1045                true,
1046            ))),
1047            false,
1048        ));
1049
1050        let l9 = Arc::new(Field::new(
1051            "l9",
1052            DataType::List(Arc::new(Field::new_list_field(
1053                DataType::Struct(Fields::from(vec![l10])),
1054                true,
1055            ))),
1056            false,
1057        ));
1058
1059        let l8 = Arc::new(Field::new(
1060            "l8",
1061            DataType::List(Arc::new(Field::new_list_field(
1062                DataType::Struct(Fields::from(vec![l9])),
1063                true,
1064            ))),
1065            false,
1066        ));
1067        let l7 = Arc::new(Field::new(
1068            "l7",
1069            DataType::List(Arc::new(Field::new_list_field(
1070                DataType::Struct(Fields::from(vec![l8])),
1071                true,
1072            ))),
1073            false,
1074        ));
1075        let l6 = Arc::new(Field::new(
1076            "l6",
1077            DataType::List(Arc::new(Field::new_list_field(
1078                DataType::Struct(Fields::from(vec![l7])),
1079                true,
1080            ))),
1081            false,
1082        ));
1083        let l5 = Arc::new(Field::new(
1084            "l5",
1085            DataType::List(Arc::new(Field::new_list_field(
1086                DataType::Struct(Fields::from(vec![l6])),
1087                true,
1088            ))),
1089            false,
1090        ));
1091        let l4 = Arc::new(Field::new(
1092            "l4",
1093            DataType::List(Arc::new(Field::new_list_field(
1094                DataType::Struct(Fields::from(vec![l5])),
1095                true,
1096            ))),
1097            false,
1098        ));
1099        let l3 = Arc::new(Field::new(
1100            "l3",
1101            DataType::List(Arc::new(Field::new_list_field(
1102                DataType::Struct(Fields::from(vec![l4])),
1103                true,
1104            ))),
1105            false,
1106        ));
1107        let l2 = Arc::new(Field::new(
1108            "l2",
1109            DataType::List(Arc::new(Field::new_list_field(
1110                DataType::Struct(Fields::from(vec![l3])),
1111                true,
1112            ))),
1113            false,
1114        ));
1115        let l1 = Arc::new(Field::new(
1116            "l1",
1117            DataType::List(Arc::new(Field::new_list_field(
1118                DataType::Struct(Fields::from(vec![l2])),
1119                true,
1120            ))),
1121            false,
1122        ));
1123
1124        let normalize_all = Schema::new(vec![l1])
1125            .normalize(".", None)
1126            .expect("valid normalization");
1127
1128        assert_eq!(normalize_all.fields().len(), 1);
1129    }
1130
1131    #[test]
1132    fn normalize_dictionary() {
1133        let a = Arc::new(Field::new("a", DataType::Utf8, true));
1134        let b = Arc::new(Field::new("b", DataType::Int64, false));
1135
1136        let one = Arc::new(Field::new(
1137            "1",
1138            DataType::Dictionary(
1139                Box::new(DataType::Int32),
1140                Box::new(DataType::Struct(Fields::from(vec![a.clone(), b.clone()]))),
1141            ),
1142            false,
1143        ));
1144
1145        let normalize_all = Schema::new(vec![one.clone()])
1146            .normalize(".", None)
1147            .expect("valid normalization");
1148
1149        let expected = Schema::new(vec![Field::new(
1150            "1",
1151            DataType::Dictionary(
1152                Box::new(DataType::Int32),
1153                Box::new(DataType::Struct(Fields::from(vec![a.clone(), b.clone()]))),
1154            ),
1155            false,
1156        )]);
1157
1158        assert_eq!(normalize_all, expected);
1159    }
1160
1161    #[test]
1162    #[should_panic(
1163        expected = "Unable to get field named \\\"nickname\\\". Valid fields: [\\\"first_name\\\", \\\"last_name\\\", \\\"address\\\", \\\"interests\\\"]"
1164    )]
1165    fn schema_field_with_name() {
1166        let schema = person_schema();
1167        assert_eq!(
1168            schema.field_with_name("first_name").unwrap().name(),
1169            "first_name"
1170        );
1171        assert_eq!(
1172            schema.field_with_name("last_name").unwrap().name(),
1173            "last_name"
1174        );
1175        schema.field_with_name("nickname").unwrap();
1176    }
1177
1178    #[test]
1179    fn schema_field_with_dict_id() {
1180        let schema = person_schema();
1181
1182        #[allow(deprecated)]
1183        let fields_dict_123: Vec<_> = schema
1184            .fields_with_dict_id(123)
1185            .iter()
1186            .map(|f| f.name())
1187            .collect();
1188        assert_eq!(fields_dict_123, vec!["interests"]);
1189
1190        #[allow(deprecated)]
1191        let is_empty = schema.fields_with_dict_id(456).is_empty();
1192        assert!(is_empty);
1193    }
1194
1195    fn person_schema() -> Schema {
1196        let kv_array = [("k".to_string(), "v".to_string())];
1197        let field_metadata: HashMap<String, String> = kv_array.iter().cloned().collect();
1198        let first_name =
1199            Field::new("first_name", DataType::Utf8, false).with_metadata(field_metadata);
1200
1201        Schema::new(vec![
1202            first_name,
1203            Field::new("last_name", DataType::Utf8, false),
1204            Field::new(
1205                "address",
1206                DataType::Struct(Fields::from(vec![
1207                    Field::new("street", DataType::Utf8, false),
1208                    Field::new("zip", DataType::UInt16, false),
1209                ])),
1210                false,
1211            ),
1212            #[allow(deprecated)]
1213            Field::new_dict(
1214                "interests",
1215                DataType::Dictionary(Box::new(DataType::Int32), Box::new(DataType::Utf8)),
1216                true,
1217                123,
1218                true,
1219            ),
1220        ])
1221    }
1222
1223    #[test]
1224    fn test_try_merge_field_with_metadata() {
1225        // 1. Different values for the same key should cause error.
1226        let metadata1: HashMap<String, String> = [("foo".to_string(), "bar".to_string())]
1227            .iter()
1228            .cloned()
1229            .collect();
1230        let f1 = Field::new("first_name", DataType::Utf8, false).with_metadata(metadata1);
1231
1232        let metadata2: HashMap<String, String> = [("foo".to_string(), "baz".to_string())]
1233            .iter()
1234            .cloned()
1235            .collect();
1236        let f2 = Field::new("first_name", DataType::Utf8, false).with_metadata(metadata2);
1237
1238        assert!(Schema::try_merge(vec![Schema::new(vec![f1]), Schema::new(vec![f2])]).is_err());
1239
1240        // 2. None + Some
1241        let mut f1 = Field::new("first_name", DataType::Utf8, false);
1242        let metadata2: HashMap<String, String> = [("missing".to_string(), "value".to_string())]
1243            .iter()
1244            .cloned()
1245            .collect();
1246        let f2 = Field::new("first_name", DataType::Utf8, false).with_metadata(metadata2);
1247
1248        assert!(f1.try_merge(&f2).is_ok());
1249        assert!(!f1.metadata().is_empty());
1250        assert_eq!(f1.metadata(), f2.metadata());
1251
1252        // 3. Some + Some
1253        let mut f1 = Field::new("first_name", DataType::Utf8, false).with_metadata(
1254            [("foo".to_string(), "bar".to_string())]
1255                .iter()
1256                .cloned()
1257                .collect(),
1258        );
1259        let f2 = Field::new("first_name", DataType::Utf8, false).with_metadata(
1260            [("foo2".to_string(), "bar2".to_string())]
1261                .iter()
1262                .cloned()
1263                .collect(),
1264        );
1265
1266        assert!(f1.try_merge(&f2).is_ok());
1267        assert!(!f1.metadata().is_empty());
1268        assert_eq!(
1269            f1.metadata().clone(),
1270            [
1271                ("foo".to_string(), "bar".to_string()),
1272                ("foo2".to_string(), "bar2".to_string())
1273            ]
1274            .iter()
1275            .cloned()
1276            .collect()
1277        );
1278
1279        // 4. Some + None.
1280        let mut f1 = Field::new("first_name", DataType::Utf8, false).with_metadata(
1281            [("foo".to_string(), "bar".to_string())]
1282                .iter()
1283                .cloned()
1284                .collect(),
1285        );
1286        let f2 = Field::new("first_name", DataType::Utf8, false);
1287        assert!(f1.try_merge(&f2).is_ok());
1288        assert!(!f1.metadata().is_empty());
1289        assert_eq!(
1290            f1.metadata().clone(),
1291            [("foo".to_string(), "bar".to_string())]
1292                .iter()
1293                .cloned()
1294                .collect()
1295        );
1296
1297        // 5. None + None.
1298        let mut f1 = Field::new("first_name", DataType::Utf8, false);
1299        let f2 = Field::new("first_name", DataType::Utf8, false);
1300        assert!(f1.try_merge(&f2).is_ok());
1301        assert!(f1.metadata().is_empty());
1302    }
1303
1304    #[test]
1305    fn test_schema_merge() {
1306        let merged = Schema::try_merge(vec![
1307            Schema::new(vec![
1308                Field::new("first_name", DataType::Utf8, false),
1309                Field::new("last_name", DataType::Utf8, false),
1310                Field::new(
1311                    "address",
1312                    DataType::Struct(vec![Field::new("zip", DataType::UInt16, false)].into()),
1313                    false,
1314                ),
1315            ]),
1316            Schema::new_with_metadata(
1317                vec![
1318                    // nullable merge
1319                    Field::new("last_name", DataType::Utf8, true),
1320                    Field::new(
1321                        "address",
1322                        DataType::Struct(Fields::from(vec![
1323                            // add new nested field
1324                            Field::new("street", DataType::Utf8, false),
1325                            // nullable merge on nested field
1326                            Field::new("zip", DataType::UInt16, true),
1327                        ])),
1328                        false,
1329                    ),
1330                    // new field
1331                    Field::new("number", DataType::Utf8, true),
1332                ],
1333                [("foo".to_string(), "bar".to_string())]
1334                    .iter()
1335                    .cloned()
1336                    .collect::<HashMap<String, String>>(),
1337            ),
1338        ])
1339        .unwrap();
1340
1341        assert_eq!(
1342            merged,
1343            Schema::new_with_metadata(
1344                vec![
1345                    Field::new("first_name", DataType::Utf8, false),
1346                    Field::new("last_name", DataType::Utf8, true),
1347                    Field::new(
1348                        "address",
1349                        DataType::Struct(Fields::from(vec![
1350                            Field::new("zip", DataType::UInt16, true),
1351                            Field::new("street", DataType::Utf8, false),
1352                        ])),
1353                        false,
1354                    ),
1355                    Field::new("number", DataType::Utf8, true),
1356                ],
1357                [("foo".to_string(), "bar".to_string())]
1358                    .iter()
1359                    .cloned()
1360                    .collect::<HashMap<String, String>>()
1361            )
1362        );
1363
1364        // support merge union fields
1365        assert_eq!(
1366            Schema::try_merge(vec![
1367                Schema::new(vec![Field::new_union(
1368                    "c1",
1369                    vec![0, 1],
1370                    vec![
1371                        Field::new("c11", DataType::Utf8, true),
1372                        Field::new("c12", DataType::Utf8, true),
1373                    ],
1374                    UnionMode::Dense
1375                ),]),
1376                Schema::new(vec![Field::new_union(
1377                    "c1",
1378                    vec![1, 2],
1379                    vec![
1380                        Field::new("c12", DataType::Utf8, true),
1381                        Field::new("c13", DataType::Time64(TimeUnit::Second), true),
1382                    ],
1383                    UnionMode::Dense
1384                ),])
1385            ])
1386            .unwrap(),
1387            Schema::new(vec![Field::new_union(
1388                "c1",
1389                vec![0, 1, 2],
1390                vec![
1391                    Field::new("c11", DataType::Utf8, true),
1392                    Field::new("c12", DataType::Utf8, true),
1393                    Field::new("c13", DataType::Time64(TimeUnit::Second), true),
1394                ],
1395                UnionMode::Dense
1396            ),]),
1397        );
1398
1399        // incompatible field should throw error
1400        assert!(Schema::try_merge(vec![
1401            Schema::new(vec![
1402                Field::new("first_name", DataType::Utf8, false),
1403                Field::new("last_name", DataType::Utf8, false),
1404            ]),
1405            Schema::new(vec![Field::new("last_name", DataType::Int64, false),])
1406        ])
1407        .is_err());
1408
1409        // incompatible metadata should throw error
1410        let res = Schema::try_merge(vec![
1411            Schema::new_with_metadata(
1412                vec![Field::new("first_name", DataType::Utf8, false)],
1413                [("foo".to_string(), "bar".to_string())]
1414                    .iter()
1415                    .cloned()
1416                    .collect::<HashMap<String, String>>(),
1417            ),
1418            Schema::new_with_metadata(
1419                vec![Field::new("last_name", DataType::Utf8, false)],
1420                [("foo".to_string(), "baz".to_string())]
1421                    .iter()
1422                    .cloned()
1423                    .collect::<HashMap<String, String>>(),
1424            ),
1425        ])
1426        .unwrap_err();
1427
1428        let expected = "Fail to merge schema due to conflicting metadata. Key 'foo' has different values 'bar' and 'baz'";
1429        assert!(
1430            res.to_string().contains(expected),
1431            "Could not find expected string '{expected}' in '{res}'"
1432        );
1433    }
1434
1435    #[test]
1436    fn test_schema_builder_change_field() {
1437        let mut builder = SchemaBuilder::new();
1438        builder.push(Field::new("a", DataType::Int32, false));
1439        builder.push(Field::new("b", DataType::Utf8, false));
1440        *builder.field_mut(1) = Arc::new(Field::new("c", DataType::Int32, false));
1441        assert_eq!(
1442            builder.fields,
1443            vec![
1444                Arc::new(Field::new("a", DataType::Int32, false)),
1445                Arc::new(Field::new("c", DataType::Int32, false))
1446            ]
1447        );
1448    }
1449
1450    #[test]
1451    fn test_schema_builder_reverse() {
1452        let mut builder = SchemaBuilder::new();
1453        builder.push(Field::new("a", DataType::Int32, false));
1454        builder.push(Field::new("b", DataType::Utf8, true));
1455        builder.reverse();
1456        assert_eq!(
1457            builder.fields,
1458            vec![
1459                Arc::new(Field::new("b", DataType::Utf8, true)),
1460                Arc::new(Field::new("a", DataType::Int32, false))
1461            ]
1462        );
1463    }
1464
1465    #[test]
1466    fn test_schema_builder_metadata() {
1467        let mut metadata = HashMap::with_capacity(1);
1468        metadata.insert("key".to_string(), "value".to_string());
1469
1470        let fields = vec![Field::new("test", DataType::Int8, true)];
1471        let mut builder: SchemaBuilder = Schema::new(fields).with_metadata(metadata).into();
1472        builder.metadata_mut().insert("k".into(), "v".into());
1473        let out = builder.finish();
1474        assert_eq!(out.metadata.len(), 2);
1475        assert_eq!(out.metadata["k"], "v");
1476        assert_eq!(out.metadata["key"], "value");
1477    }
1478}
arrow_schema/schema.rs

arrow_schema/
schema.rs
